The AI race is on a hot streak and nowhere more so than in the hypercompetitive technology and financial markets. This level of competition is exactly what’s needed to generate profound innovation. Customers reward the best companies and models, and less effective and popular competitors fall behind quickly and ruthlessly.
Technology companies from Meta to Amazon to Alphabet are making major investments to remain relevant in the AI race, with more than $660 billion in total AI spending planned for 2026. Microsoft’s investments in particular are eye-popping—including $37.5 billion in its last financial quarter.
However, Wall Street is now demanding evidence of product uptake and pathways to profitability—and Microsoft is stumbling. The company’s latest earnings report led to a large drop in share prices, as investors and analysts raised concerns about its massive spending on AI infrastructure without the kinds of tangible returns that a really valuable product should demonstrate.
It’s simple: very few people want to use Copilot and almost nobody wants to pay for it.
Only a small percentage of Microsoft customers use the product, and only 3.3 percent of Microsoft 365 and Office 365 customers are actually paying for it. In fact, Copilot subscribers are increasingly choosing competing options, like ChatGPT and Gemini. By all standard measures, Microsoft’s Copilot AI agent should be getting lapped by the competition.
Instead of building better products, Microsoft has resorted to its decades-old playbook of forcing a product onto locked-in customers, in this case pushing customers towards more expensive integrated suites with its AI functions bundled in—functions that the customers were not freely choosing to purchase. That’s not competition—it’s a bald assertion of market power that undermines competition.
Just last month, the company announced it would be ending its standalone SharePoint Online and OneDrive for Business plans, requiring customers already locked-in to move to more expensive bundled Microsoft 365 suites that include Copilot. These bundled plans cost anywhere from 5 to 10 times more than the individual subscriptions. And, adding insult to injury, Microsoft is also raising prices by 5.5 to 8.3 percent later this year. For users of the standalone products, this change increases costs considerably—all for products they don’t use or want.
This follows Microsoft adding Copilot to Microsoft 365 in Australia and several Southeast Asian countries and then raising prices, again imposing its AI agent on customers and making them pay for a bundle that has no technology justification. Australia sued Microsoft last year, accusing the company of misleading its users into paying higher prices for the products bundled together and of concealing the cheaper, Copilot-free plan until users began the cancellation process. The case is still ongoing, although Microsoft conceded the deception and offered refunds to nearly 3 million Australians.
And just this week, recognizing that shoehorning Copilot into products won’t prevent unimpressed users from sticking with ChatGPT, Microsoft has been vocal about the dangers of employees using AI outside of company systems, a practice known as “shadow AI.” While unapproved AI use can certainly be a security risk, it’s also no coincidence that Microsoft—the company with the workplace AI product no one wants to use—is so outspoken against workplace AI choice.
Forcing or tricking users into adopting a product when there are better alternatives isn’t just an exploitative business strategy. It’s also a signal of weakness: markets are functioning well when customers seek it out and choose the product that they value the most. Having spent billions of dollars to build a sub-standard product that customers don’t want, Microsoft has taken a desperate path. That unfortunate decision shouldn’t be allowed to distort the most important competitive markets in technology in decades.
